Removed gamess parser

.ocamlinit

@@ -7,22 +7,10 @@ let () =
 #require "lacaml";;
 #directory "_build";;
 #directory "_build/Utils";;
+#directory "_build/Nuclei";;
 #directory "_build/Basis";;
 #directory "_build/HartreeFock";;
 
-#load "Powers.cmo";;
-#load "Zkey.cmo";;
-#load "AngularMomentum.cmo";;
-#load "Constants.cmo";;
-#load "Util.cmo";;
-#load "Coordinate.cmo";;
-#load "Zmap.cmo";;
-#load "Zkey.cmo";;
-#load "ContractedShell.cmo";;
-#load "ShellPair.cmo";;
-#load "TwoElectronRR.cmo";;
-#load "ERI.cmo";;
-
 #use "topfind";;
 
 

Basis/Basis.ml

@@ -65,9 +65,9 @@ let to_string b =
 
 
 
-let of_nuclei_and_basis_filename ~nuclei ~filename = 
+let of_nuclei_and_basis_filename ~nuclei filename = 
   let general_basis = 
-    GamessReader.read ~filename
+    GeneralBasis.read filename
   in
   of_nuclei_and_general_basis nuclei general_basis
 

Basis/Basis.mli

@@ -11,7 +11,7 @@ val to_string : t -> string
 (** Pretty prints the basis set in a string. *)
 
 
-val of_nuclei_and_general_basis : Nuclei.t -> GeneralBasis.t list -> t
+val of_nuclei_and_general_basis : Nuclei.t -> GeneralBasis.t -> t
 (** Takes an array of {!Nuclei.t}, and a {!GeneralBasis.t} (such as cc-pVDZ
   for instance) and creates the corresponding atomic basis set.
   All the {!Element.t}s of the array of {!Nuclei.t} are searched in 
@@ -22,7 +22,7 @@ val of_nuclei_and_general_basis : Nuclei.t -> GeneralBasis.t list -> t
 
 
 
-val of_nuclei_and_basis_filename : nuclei:Nuclei.t -> filename:string -> t
+val of_nuclei_and_basis_filename : nuclei:Nuclei.t -> string -> t
 (** Same as {!of_nuclei_and_general_basis}, but taking the {!GeneralBasis.t}
   from a file. 
   *)

Basis/BasisLexer.mll

@@ -1,33 +0,0 @@
-{
-exception SyntaxError of string
-
-open GamessParser
-
-}
-
-let eol = ['\n']
-let white = [' ' '\t']+
-let element   = ['A'-'Z' 'a'-'z']+ white? eol
-let ang_mom  = ['S' 'P' 'D' 'F' 'G' 'H' 'I' 'J' 'K' 'L' 'M' 'N' 'O' 
-                's' 'p' 'd' 'f' 'g' 'h' 'i' 'j' 'k' 'l' 'm' 'n' 'o' ]
-               white
-let integer  = ['0'-'9']+ 
-let real = '-'? integer '.' integer (['e' 'E'] ('+'|'-')? integer)? 
-
-
-rule read_all_rule  = parse
-  | eol            { EOL }
-  | white          { SPACE }
-  | ang_mom as a   { ANG_MOM (a.[0]) }
-  | element as e   { ELEMENT (String.trim e) }
-  | integer as i   { INTEGER (int_of_string i) }
-  | real    as f   { FLOAT (float_of_string f) }
-  | eof            { EOF }
-
-
-{
-  let rec read_all lexbuf =
-     match read_all_rule lexbuf with
-     | SPACE -> read_all_rule lexbuf
-     | x -> x
-}

Basis/GamessParser.mly

@@ -1,49 +0,0 @@
-/* Parses basis sets GAMESS format */
-
-%{
-
-%}
-
-%token <string> ELEMENT
-%token <char> ANG_MOM
-%token <int> INTEGER
-%token <float> FLOAT
-%token SPACE
-%token EOL
-%token EOF
-
-%start input
-%type <GeneralBasis.t> input
-
-%% /* Grammar rules and actions follow */
-
-input:
-  | basis { $1 }
-  | EOL input { $2 }
-
-basis:
-  | element shell_array EOL { ($1, $2) }
-  | element shell_array EOF { ($1, $2) }
-
-element:
-  | ELEMENT { Element.of_string $1 }
-
-shell_array:
-  | shell_list { Array.of_list @@ List.rev $1 }
-
-shell_list:
-  | { [] }
-  | shell_list shell { $2 :: $1 }
-
-shell:
-  | ANG_MOM INTEGER EOL primitive_list { (AngularMomentum.of_char $1, Array.of_list @@ List.rev $4 ) } 
-
-primitive_list: 
-  | { [] }
-  | primitive_list primitive { $2 :: $1 }
-
-primitive: 
-  | INTEGER FLOAT FLOAT EOL { GeneralBasis.{exponent=$2 ; coefficient=$3 } }
-
-
-

Basis/GamessReader.ml

@@ -1,18 +0,0 @@
-(** Read a basis set file in GAMESS format and return an association list where the key is an
-    Element.t and the value is the parsed basis set. *)
-let read ~filename =
-  let lexbuf =
-    let ic = open_in filename in
-    Lexing.from_channel ic
-  in
-  let rec aux accu = 
-    try
-      let key, basis =
-        GamessParser.input BasisLexer.read_all lexbuf
-      in
-      aux ((key, basis)::accu)
-    with
-    | Parsing.Parse_error -> List.rev accu
-  in
-  aux []
-

Basis/GeneralBasis.ml

@@ -1,16 +1,84 @@
 (** General basis set read from a file  *)
-type primitive = {
-    exponent: float ;
-    coefficient: float
-    }
+
+type primitive =
+  {
+    exponent   : float ;
+    coefficient: float ;
+  }
 
 type general_contracted_shell = AngularMomentum.t * (primitive array)
 
-type t = Element.t * (general_contracted_shell array)
+type element_basis = Element.t * (general_contracted_shell array)
+
+type t = element_basis list
 
 
-module Am = AngularMomentum
+exception No_shell
+exception Malformed_shell of string
 
+
+let read_shell ic =
+  try
+
+    let shell, n =
+      try
+        Scanf.sscanf (input_line ic) " %c %d " (fun shell n -> shell, n)
+      with
+      | End_of_file -> raise No_shell
+      | Scanf.Scan_failure m -> raise (Malformed_shell m)
+    in
+
+    let rec loop = function
+      | 0 -> []
+      | i -> let contraction = 
+               let line = (input_line ic) in
+               try Scanf.sscanf line " %_d %f %f "
+                     (fun exponent coefficient -> { exponent ; coefficient })
+               with _ -> raise (Malformed_shell (Printf.sprintf
+                     "Expected %d %c contractions, error at contraction %d:\n%s"
+                      n shell (n-i+1) line))
+        in
+        contraction :: loop (pred i)
+    in
+    Some (AngularMomentum.of_char shell, Array.of_list (loop n))
+
+  with
+  | No_shell -> None
+
+
+
+
+let read_element ic =
+  try
+
+    let element =
+      Scanf.sscanf (input_line ic) " %s " Element.of_string
+    in
+
+    let rec loop () =
+      match read_shell ic with
+      | Some shell -> shell :: loop ()
+      | None -> []
+    in
+    Some (element, Array.of_list (loop ()) )
+
+  with
+  | Element.ElementError _ -> None
+  | End_of_file -> None
+  
+
+
+let read filename =
+  let ic = open_in filename in
+  let rec loop () =
+    match read_element ic with
+    | Some e -> e :: loop ()
+    | None -> []
+  in
+  loop ()
+
+
+  
 let string_of_primitive ?id prim =
   match id with
   | None -> (string_of_float prim.exponent)^" "^(string_of_float prim.coefficient)
@@ -22,7 +90,7 @@ let string_of_contracted_shell (angular_momentum, prim_array) =
     Array.length prim_array
   in
   Printf.sprintf "%s %d\n%s"
-    (Am.to_string angular_momentum) n
+    (AngularMomentum.to_string angular_momentum) n
     (Array.init n (fun i -> string_of_primitive ~id:(i+1) prim_array.(i))
      |> Array.to_list
      |> String.concat "\n")

Basis/GeneralBasis.mli

@@ -0,0 +1,108 @@
+(** General basis set read from a file.
+
+Basis set files are stored in GAMESS format, for example:
+{[
+HYDROGEN
+S   3
+  1     13.0100000              0.0196850
+  2      1.9620000              0.1379770
+  3      0.4446000              0.4781480
+S   1
+  1      0.1220000              1.0000000
+P   1
+  1      0.7270000              1.0000000
+
+HELIUM
+S   3
+  1     38.3600000              0.0238090
+  2      5.7700000              0.1548910
+  3      1.2400000              0.4699870
+S   1
+  1      0.2976000              1.0000000
+P   1
+  1      1.2750000              1.0000000
+]}
+
+When building the atomic basis set of a molecular system, the basis functions
+are created by picking the data from the general basis set. This data structure
+simplifies the creation of the atomic basis set.
+*)
+
+type primitive = private
+  {
+    exponent   : float ;
+    coefficient: float ;
+  }
+
+type general_contracted_shell = AngularMomentum.t * (primitive array)
+
+type element_basis = Element.t * (general_contracted_shell array)
+
+type t = element_basis list
+
+
+exception No_shell
+exception Malformed_shell of string
+
+
+val read : string -> t
+(** Reads a basis set file and return an association list where
+    the key is an {!Element.t} and the value is the parsed basis set.
+*)
+
+
+val read_element : in_channel -> element_basis option
+(** Reads an element from the input channel [ic]. For example,
+{[
+HYDROGEN
+S   3
+  1     13.0100000              0.0196850
+  2      1.9620000              0.1379770
+  3      0.4446000              0.4781480
+S   1
+  1      0.1220000              1.0000000
+P   1
+  1      0.7270000              1.0000000
+
+}]
+returns
+{[
+Some
+ (Element.H,
+  [(AngularMomentum.S,
+    [{exponent = 13.01; coefficient = 0.019685};
+     {exponent = 1.962; coefficient = 0.137977};
+     {exponent = 0.4446; coefficient = 0.478148}]);
+   (AngularMomentum.S, [{exponent = 0.122; coefficient = 1.}]);
+   (AngularMomentum.P, [{exponent = 0.727; coefficient = 1.}])])
+]}
+
+@raise Malformed_shell if the input is not well formed.
+*)
+  
+
+val read_shell : in_channel -> general_contracted_shell option
+(** Reads a shell from the input channel [ic]. For example,
+{[
+S   3
+  1     13.0100000              0.0196850
+  2      1.9620000              0.1379770
+  3      0.4446000              0.4781480
+]}
+returns
+{[
+Some
+ (AngularMomentum.S,
+  [{exponent = 13.01 ; coefficient = 0.019685};
+   {exponent = 1.962 ; coefficient = 0.137977};
+   {exponent = 0.4446; coefficient = 0.478148}])
+]}
+
+@raise Malformed_shell if the input is not well formed.
+*)
+
+
+val to_string : string * (general_contracted_shell array) -> string
+(** Pretty-prints the basis set of an {Element.t}. *)
+
+

Simulation.ml

@@ -37,12 +37,12 @@ let make ?multiplicity:(multiplicity=1) ?charge:(charge=0) ~nuclei basis =
   }
 
 
-let of_filenames ?multiplicity:(multiplicity=1) ?charge:(charge=0) ~nuclei basis =
+let of_filenames ?multiplicity:(multiplicity=1) ?charge:(charge=0) ~nuclei basis_filename =
   let nuclei =
     Nuclei.of_filename nuclei
   in
   let basis =
-    Basis.of_nuclei_and_basis_filename ~nuclei ~filename:basis
+    Basis.of_nuclei_and_basis_filename ~nuclei basis_filename
   in
   make ~charge ~multiplicity ~nuclei basis